Tungsten Replicator 2.0.7 is released

Tungsten Replicator 2.0.7 was released today. In addition to a large number of bug fixes, this release adds several improvements for multi-master management, and support for Amazon RDS (as a slave).

While the Release Notes show a long list of improvements, I would like to focus on some of them that improve the handling of multi-master deployments.

When we released version 2.0.6, we added the first revision of the cookbook recipes in the build. That was still a green addition, which caused several bug reports. But since then, we have integrated the cookbook in our internal testing, making these recipes more robust and reliable. We are also planning to improve them and eventually integrate all of them into the main installer. Probably in the next release. In the meantime, I would like to point out the main additions:

• The clear_* scripts now ask for confirmation, or use an environment variable, if set. This reduce the risk of accidental data loss.
• The bootstrap.sh script makes a better job of setting the environment;
• You have scripts that add nodes to an existing master/slave or star topology;
• There are scripts to install native MySQL replication and show how to take over with Tungsten;
• There are scripts that perform a switch between master and slaves in master/slave topology;
• There is also a script that installs all topologies, one by one, and tests that they replicate correctly;
• Finally, there is a script that collect logs from all servers into a single bundle. This is useful when you want to diagnose a problem.
• Tungsten Replicator can now work with MySQL 5.6. There are cases where you would need to re-start the MySQL server with a new option (native CRC conflicts with current understanding of the binlog by Tungsten), but it can be installed. Also Tungsten Sandbox can get around that limitation on its own.

Another useful addition is a feature that I have recently presented at FOSDEM 2013. It's an enhancement of Tungsten ability of preventing conflicts in multiple masters deployments. It is now possible to whitelist sources, so that they will be updated by every master, and it is possible to define in greater detail which schemas can be updated by which sources.

Finally, we have made some great advancements in the way Tungsten can replicate from MySQL to Oracle servers. It is now possible to bypass some previously heavy manual operations, when we needed to recreate DDL statements manually on the Oracle side. Now there is a tool that can automate most of the DDL creation. My colleague Linas will explain that in his blog soon.

There are many reasons for trying out the new version. You can download it from Tungsten site.